Perforated Fairings for Landing Gear Noise

Control

K. Boorsma ∗ and X. Zhang†

University of Southampton, UK

N. Molin‡

Airbus, Toulouse, France

I. Abstract

Landing gears of commercial aircraft make an important contribution to total aircraft noisein the approach configuration. Using fairings to shield components from high speed impingementreduces noise. Furthermore, perforating these fairings has been confirmed by flight tests to furtherenable noise reduction. Following a more fundamental study on the application of perforatedfairings, a study has been performed to investigate and optimize the benefits of bleeding airthrough landing gear fairings. By means of wind tunnel tests, an aerodynamic and acousticsurvey has been performed on a simplified A340 look alike Main Landing Gear to explore theinfluence of (perforated) fairings on the lower part of the gear.

The results show that for this specific case, the application of impermeable fairings reducesnoise in the mid- and high frequency range by shielding sharp edged components from high velocityimpingement. However, below 1 kHz the noise is shown to increase significantly. Application ofthe perforations is shown to diminish this low frequency increase whilst maintaining the reductionin the mid- and high frequency range. The aerodynamic and acoustic measurements point in thedirection of the separated flow of the fairings interacting with the downstream gear componentsresponsible for the low frequency noise increase. Bleeding of the air through the fairings reducesthe large scale turbulence in the proximity of these components and hence diminishes the lowfrequency increase. A sharp noise increase is present in the ground view direction at the 500Hz centred tertsband for the solid fairing, which is confirmed to originate from the articulationlink fairing. The flow around the wheels remains largely unaltered by application of the fairings.Furthermore the application of fairings is shown to suppress vortex shedding from the lower partof the main leg.

A study on the effectiveness of perforation location reveals that the stagnation area perforationsare responsible for most of the low frequency noise decrease. Both articulation link and undertrayfairing contribute to the reduction. Exposing the perforate outside the stagnation area inducesperforate self-noise, most significantly on the sides of the lower articulation link fairing. Agreeingwith previous studies, the phenomenon scales with the local component of the shearing flow velocityand orifice diameter.

II. Introduction

A steady increase in the number of flights and airports has initiated research into aviation noisefor commercial aircraft. A more stringent legislation and increased environmental awareness on thissubject has led to an extra push. Research started with tackling the most dominant component,engine noise. As a result, engine noise has been reduced dramatically over the past decades,

∗PhD Student, SES, University of Southampton, SO17 1BJ, UK.†Professor, SES, University of Southampton, SO17 1BJ, UK, Associate Fellow AIAA.‡Engineer, EEA Department, 316 Route de Bayonne, PO Box M0112/4, 31060 Toulouse, Cedex 03, France.

1 of 20

American Institute of Aeronautics and Astronautics

thereby increasing relative importance of airframe noise for noise certification and environmentalconsiderations.

Landing gear noise is the dominant component of airframe noise, especially for larger aircraft.1

Purely designed to perform their main function, landing gears feature an extremely aerodynamicunfriendly shape. The interaction of airflow with protrusions and cavities gives rise to unsteadyflow phenomena constituting a potent sound generating mechanism.

Shielding landing gear components with fairings and thereby preventing this interaction hasbeen considered before.1,2 High speed flow deflection onto other components and fairing self-noiseare disadvantages of this measure. Bleeding air through the fairing results in a redistribution ofthe air flow. Flight tests using perforated fairings have been conducted before3 and demonstrateda noise decrease. The aim of the present research is to investigate and optimize the noise reductionpotential by perforating these fairings.

The effect of bleeding air through a fairing on aeroacoustic performance of landing gear is arelatively new topic. Recent research on the subject4,5 has confirmed the flight test results, al-though the corresponding flow behavior has not been investigated. A previous study on perforatedfairings employing a simplified bluff body geometry6 has shown that breakdown of the large scaleflow structures associated with the fairings can be a major benefit, although one should be care-ful to prevent the adverse effect of perforate self-noise. Armed with the lessons learnt from thisexperiment, aerodynamics and acoustics of a simplified landing gear model have been investigated.

III. Model and test set-up

A visualization of the landing gear model is shown in Figure 1(a). It is a simplified 0.25

(a) Gear components (2 wheels omitted for clarity) (b) Model as suspended in the tunnel

Figure 1. Visualization of the landing gear model.

scale model, based on the A340 Main Landing Gear. Any landing gear parts located above thearticulation link are omitted, since the investigation concentrates on the articulation link andundertray fairing shielding articulation link and bogie beam respectively. Details on articulation,torque link and brake discs and links are accurately represented, although hoses are not present.Further more the outside rim of the wheels are covered with hub caps, and tyre thread is omitted.

Figure 1(b) shows the main strut inclined perpendicular to the flow and the tunnel side wall.A coordinate system can be defined originating in the center of the main strut-bogie beam junc-tion. The x-coordinate is in streamwise direction, positive pointing downstream. The y- andz-coordinate are defined in Figure 1(b).

The blue coloured part of the fairings are perforated with round holes resulting in an openarea of 40%. To discover the influence perforate location, tape has been applied to cover variousparts of the perforated fairings. The three main configurations tested are without fairings, solid

2 of 20

American Institute of Aeronautics and Astronautics

fairings (perforations taped) and perforated fairings indicated by LG, Solid and Perf respectivelyin the Figures.

Flow and acoustic measurements were performed in the closed circuit University of Southamp-ton 7′ × 5′ wind tunnel, using the 2.1 m wide, 1.7 m high octagonal cross section as depictedin Figure 1(b). The blockage amounts to 10% based on model frontal area (with fairings on).Freestream turbulence levels are approximately 0.2% over the velocity range considered, boundedby the maximum tunnel speed of U∞ = 40 m/s. To monitor velocity scaling of aerodynamicsand acoustics, tests are performed at U∞ = 20, U∞ = 30 and U∞ = 40 m/s. An aircraft ap-proach speed of 80 m/s means that frequencies in the model scale test at U∞ = 40 m/s will beapproximately twice the values of the ones for full scale flight.

For the aerodynamic tests, 3 different configuration have been investigated; the plain landinggear (no fairings), fairings applied with perforations taped and fairings applied with all perforationsexposed indicated respectively by LG, Solid and Perf in the Figures.

IV. Experimental apparatus

IV.A. Oil flow visualization

A mixture of titanium dioxide, paraffin and oleic acid was applied to the surfaces of the shell andstrut. The oil is transported along surface streaklines leaving an image of the time-averaged flowfield after it has dried up. Flow direction and separation are typical phenomenon observed withthis technique.7

IV.B. Particle Image Velocimetry

Particle Image Velocimetry (PIV) was used to obtain information about the velocity field in thexz-plane at the model center (y=0). Measurements were performed using a Dantec FlowMapsystem. A 120 mJ Nd:YAG dual-cavity laser was fixed on a table to the side of the 7′× 5′ tunnel,illuminating the relevant plane by firing through a small hole in the tunnel sidewall. A water-based seed generator was used to seed the flow. An 80C60 HiSense CCD camera (1280×1024 pixelresolution) was mounted above the tunnel, looking through a glass ceiling plate perpendicular tothe laser sheet. The camera and laser were traversed in order to visualize the flow field upstreamof the articulation link and the near wake downstream of the torque link and aft wheel. Using a60 mm lens, the image size of the planes was around 120 × 100 mm. This resulted in 2 traversepositions of the camera to visualize the articulation link area and 7 traverse positions to visualizethe near wake area with a decent overlap of around 15%. 500 image pairs were recorded perconfiguration. Time between recordings was kept constant at 0.5 s. The time between each pulsevaried with tunnel speed in the order of 50 µs.

The images were post-processed using adaptive correlation with 2 refinement steps, startingwith 128×128 and ending with 32×32 pixel interrogation area size. Vertical and horizontal overlapof the 2 images in each pair is set to 75%. Erroneous vectors were removed using a range andpeak validation.

IV.B.1. Uncertainty

The accuracy of the instantaneous velocity fields can be estimated by assuming an accuracy in thecorrelation of 0.1 pixel displacement.8 This corresponds to a maximum error in the velocity of 0.40m/s. Using error analysis for multi-sample experiments as described by Moffat,9 the uncertaintyin a time-averaged vector is 0.02 m/s.

IV.C. Pressure tappings

Static pressure was measured at various positions on the wheel surface. The centerline was tappedat 5 ◦ increment on a 90 ◦ section. Hence the wheel is rotated in 4 steps in order to get a 360 ◦

centerline pressure distribution. Complimentary to the centerline tappings, the shoulder of the

3 of 20

American Institute of Aeronautics and Astronautics

tyre surface was instrumented with 32 taps. The rotation of the wheel in steps of 90 ◦ incrementyields 4 locations for measurement of the shoulder taps. Since the model is inclined parallel to thestreamwise xz-plane, left and right wheels should yield identical pressure distributions. Thereforeonly the pressures on the front and rear wheels located in the half space with positive y-coordinate(see Figure 1) are measured.

The total of 51 taps are connected to a Scanivalve ZOC22B/32PxX2 - 20’ H20 pressure scannervia tubing inside the wheel. The ZOC module inside the wheel contains piezoresistive pressuresensors (capable of measuring up to 64 channels) and a pressure scanner temperature sensor. Thisdata is then digitized in the RADBASE3200 and passed to the PC for temperature correctedunit conversion. Using the internal calibration, the system readily gives a pressure output and anindividual calibration is not deemed necessary.

The normalized pressure coefficient can be defined as

Cp =p− p∞

0.5ρU∞2 , (1)

where p∞ is the static pressure of the tunnel pitot tube and ρ is derived from barometer andtemperature readings in the tunnel using the ideal gas law. Since the reference pressure of therecording instrumentation is connected to the static pressure port of the tunnel pitot tube, theoutput is a measure for p− p∞.

In each frame, the channels are sampled at a rate of 50 µs and averaged over 32 samples.For each run, 50 frames of data are acquired and averaged to obtain time averaged pressure. Tominimize the influence of drifting of the signal, a zero run was performed before each measurement,which was subtracted before conversion to Cp values.

IV.C.1. Uncertainty

The error estimation by Moffat9 is used, which distinguishes fixed and variable sources, dependingon whether the error it introduces is steady or changes during the time of one complete experi-ment. The fixed error consists of the accuracy of the data acquisition system and it’s calibration.Scanivalve claims an accuracy of the current system of ±0.12% of the full scale range, whichconverts into ∆Cp = ±0.006 at U∞ = 40m/s. The variable error is estimated by comparing thevalues of the 50 frames. The average standard deviation at U∞ = 40 m/s equals 0.013, indicatingan uncertainty of 0.002 for the 50 frames. Adding the fixed and variable errors by calculating theroot-sum-square9 yields a total uncertainty of 0.006 at this velocity.

IV.D. Unsteady pressure sensors

Unsteady pressure sensors have been placed on various positions flush with the gear surface. Forthe present test, sensors have been placed in positions 1, 4 and 5 indicated in Figure 2. Additionally2 sensors have been placed on the wheel centerline, spaced ∆ϕ = 180 ◦ apart. Both front and rearwheel are rotated in 4 steps to obtain unsteady centerline pressures around the wheel circumferencespaced ∆ϕ = 45 ◦ apart using the 2 sensors.

The pressure sensors exhibit a reference tube which is connected to the static pressure fromthe tunnel pitot tube, enabling real time measurement of the fluctuating surface pressure. Thesensors used are 2.4 mm diameter Kulite XCQ-093 sensors exhibiting a range of 0.35 bar anda natural frequency of 150 kHz. They are powered by a 8 channel VISHAY model 2150 straingauge amplifier. Similar as for the microphones, analog to digital conversion was performed usinga National Instruments’ PXI-4472 24 bit data acquisition card, controlled by a PC using LabViewsoftware.

To convert the output voltage to pressure units, a calibration was performed using a DruckDPI 601-F pressure calibrator for each individual Kulite, assuming a linear relationship betweenvoltage and pressure.

To minimize the influence of drifting of the signal, a zero run was performed before eachmeasurement, which was subtracted before conversion to Cp values. The data were sampled at

4 of 20

American Institute of Aeronautics and Astronautics

Nr Locationx(mm)

y(mm)

z(mm)

1 Main leg (upper) 40 0 460

2 Main leg (lower) 40 0 213

3Bogie-artlink junc-tion

-143 0 136

4Undertray surface(facing down)

40 0 -129

5 Rear bogie cap 237 0 -154

Figure 2. Pressure sensor locations.

a frequency of f = 24 kHz and a block size of n = 16384, averaged over 100 blocks, giving aresolution of ∆f = 1.46 Hz. Equation 1 was used to convert the pressures to Cp values with inputfrom barometer and temperature readings in the tunnel documented for each run.

IV.D.1. Uncertainty

The fixed error consists of the accuracy of the sensors and the calibration. The quoted accuracyof the calibrator is ±0.05% of it’s 0.5 bar range. The typical error of the Kulites due to combinednon-linearity, hysteresis and repeatability is quoted at 0.1% of the 0.35 bar range. These combinedfigures add up to a fixed error of ∆Cp = ±0.04 at a freestream velocity of U∞ = 40 m/s.

The variable error is estimated by comparing the average values of the 100 blocks. For thetime averaged pressure, the average standard deviation at U∞ = 40 m/s equals 0.008, indicatingan uncertainty of 0.0008 for the 100 blocks. For the time averaged value of Cp at U∞ = 40 m/s,this yields a total uncertainty of ±0.04.

IV.E. Microphones

Various microphones have been placed on the model (flush with its surface). Referring to Figure2, location 1,2,3 and 5 have been instrumented.

The microphones used are Panasonic Omnidirectional Back Electret Condenser Cartridges,series WM-61A, with a response range of 20 Hz to 20 kHz. Since they are electret microphones,the signal was powered by preamplifiers, especially manufactured for this purpose. Analog todigital conversion was performed using a National Instruments’ PXI-4472 data acquisition card,controlled by a PC using LabView software. A calibration factor was calculated using the recordedsignal of a B&K pistonphone (type 4230) to convert the raw data to narrowband spectra with areference pressure pref = 2× 10−5 Pa. A flat response over the frequency range is assumed. Thisassumption proved good enough for the present experiment after determination of the transferfunction between a Panasonic and a calibrated 1/2” B&K microphone, both exposed to whitenoise.

The data were sampled at a frequency of fsample = 48 kHz and a block size of n = 16384,averaged over 60 blocks.

IV.E.1. Uncertainty

The accuracy of the microphone measurements consists of many parts and is hard to estimate.However microphone levels are predominantly compared between different configurations, thus thefixed error is unimportant.

5 of 20

American Institute of Aeronautics and Astronautics

The variable error is estimated by comparing the values of the 60 acquired blocks and adds upto 0.5 dB for both sets of microphones, roughly constant for each frequency band.

IV.F. Phased microphone array

Two similar phased arrays are positioned in the starboard side wall and the ceiling of the 7′ × 5′

wind tunnel. They both consist of 56 microphones, spirally placed in a wooden board with adiameter of 700 mm, flush with its surface. The perpendicular distance between array and model

Figure 3. Set-up of microphone arrays in the 7′ × 5′ tunnel.

center is 0.73 m for the ceiling array and 1.18 m for the side wall array. The different distanceof the arrays to the model results in different resolution of the beamforming plot. Since theaperture of both arrays is the same and resolution varies linearly with distance, the side wall arrayexhibits a 61% better resolution than the ceiling array. Powering for the arrays is identical to themicrophones as described in section IV.E. The same holds for the data acquisition, which wasdone simultaneously for both arrays. However sampling frequency is set to f = 48 kHz, block sizen = 8192 and the data were averaged over 120 blocks.

The beamforming code for the present experiments was written by Fenech10 and is basedon conventional frequency-domain beamforming. 1/3-Octave band averaged beamforming plotswere calculated for comparison between configurations, however the finite aperture of the arrayprevented sufficient resolution for localization below 2 kHz.

The side view beamforming plots from the ceiling array are aligned in the xz-plane at y=0.However, the ground view plots from the side wall array are aligned with the bogie beam androtated by 33 ◦ going through the model origin. For the previously mentioned integration of thebeamforming plots, squared pressures are summed between x = ±0.2 m and z = −0.1 m to z = 0.3m for the ceiling array and between x = −0.4 m to x = 0.3 m and y = ±0.15 m for the side wallarray.

IV.F.1. Uncertainty

The fixed error of the beamforming plots and their integrated levels is not relevant since theyare only used for a comparison between configurations. A rough estimate for the variable erroris obtained from a comparison between two identical configurations. This yields a maximumuncertainty of 1.5 dB for the beamforming plots and 0.45 dB for the integrated levels.

6 of 20

American Institute of Aeronautics and Astronautics

V. Results and discussion

The results are divided into sections. Firstly results of the aerodynamic measurements arediscussed, starting with flow around the fairings, near wake flow and flow around the wheels. Thenacoustics are highlighted including microphone array and on-surface microphone measurements.

V.A. Fairing Flow

This section describes time-averaged flow features on the fairings. A front view of oil flow visual-ization on undertray and articulation link fairing is shown in Figure 4. The undertray stagnation

1 Undertray stagnation area

2 Separation and reattachment after bend

3 Development of horseshoe vortex

4 Lower artlink stagnation line

Figure 4. Front view of flow visualization of solid fairings (U∞ = 40 m/s).

line divides the flow in upward and downward directions (z-direction). The upward deflected flowimpinges on the lower articulation link fairing and forms a horseshoe vortex around it. The lowerarticulation link itself divides the flow in transverse direction (y-direction), where it separates fromthe lower articulation sides as depicted in Figure 5. The streamlines on the upper articulation linkindicate attached flow until the transition to the main leg is reached.

PIV is used to visualize the flow directly upstream of the articulation link fairing on the modelcenterline (y=0) in Figure 6. The effect of the fairing is to deflect the flow more in upwarddirection, since it’s cross sectional frontal area is large compared to the bare articulation link. Theimpact of perforations is relatively small. Noticeable is the increase in velocity in the stagnationpoint area between z=320 mm and z=380 mm caused by the bleeding of the air. However, flowdirection pattern remains largely unaltered, indicating that more porosity is needed if reductionof upward deflected velocities is desired.

Figure 7 visualizes the ground view of the downward directed flow by the undertray. The flowremains attached to the fairing surface until the trailing edge apart from the laminar separationbubble at the sharp bend close to the stagnation line. Instead of leaving the undertray at thetrailing edge, part of the flow escapes sideways between the wheels as visualized in Figure 7. Afterreaching the undertray side, a side edge vortex forms of which the path is visible in the figure. Anunsteady pressure sensor is positioned flush with the undertray surface at location 4 (see Figure 2).The flow visualization readily revealed attached flow in this region. Hence large scale fluctuationsare not present and pressure spectra are dominated by electronic noise and therefore omitted. The

7 of 20

American Institute of Aeronautics and Astronautics

1 Separation on lower artlink side

2Transition between lower and upperartlink: separation and reattachment

3Separation and reattachment after bendon upper artlink

Figure 5. Side view of flow visualization of solid fairings (U∞ = 40 m/s).

(a) LG (b) Solid (c) Perf

Figure 6. Time-averaged velocity contours in the xz-plane at y=0 (articulation link area, U∞ = 40 m/s).

8 of 20

American Institute of Aeronautics and Astronautics

1 Separation and reattachment after bend

2Sideways bending of flow between thewheels

3 Undertray side edge vortex

Figure 7. Ground view of flow visualization of solid fairings (U∞ = 40 m/s).

perforated and solid fairing configuration yield the same result, amounting to a mean pressure ofCp = 0.22 at this location.

V.B. Near Wake Flow

The flow conditions directly downstream of the gear (torque link, main leg and rear bogie area)have been investigated using PIV and unsteady pressure sensors. Time averaged velocity contoursin the model centerplane are shown in Figure 8. As noticed before, the flow above z=340 mm inFigure 6(a) is not deflected upwards by the fairing, which results in a high streamwise velocityregion in the top of Figure 8(a). The vectors illustrate that without fairings applied, the dominantwake flow direction is streamwise. An area of relatively high speed flow (0.7U∞) emerges betweenz=-100 and z=50 mm, resulting from the flow penetrating through the gear between the bogiebeam and wheels. Application of the fairings alters the wake structure significantly. From thetorque link junction, the flow is directed in upward and downward direction. Apparently deflectionof the air in z-direction by the fairings results in spanwise flow features in the wake. Since theundertray fairing prevents the air from penetrating through the gear, the high speed velocityregion between z=-100 and z=50 mm partly disappears. Additionally, a dead flow area emergesbelow z=0 mm. The flow that manages to escape in streamwise direction from the undertay sideedge (Figure 7) results in a small area of high speed streamwise flow between z=0 and z=50 mm.

The corresponding contours of standard deviation are shown in Figure 9. As expected, appli-cation of fairings yields very low values of u′ in the dead flow area below z=0 mm. Apart fromthe dead flow area, the near wake generally exhibits lower values of u′ without fairings applied.An explanation could be that the larger size of the fairings compared to the plain gear introduceslarge scale fluctuations and therefore higher values of u′.

9 of 20

American Institute of Aeronautics and Astronautics

(a) LG (b) Solid

Figure 8. Time-averaged velocity contours in the xz-plane at y=0 (torque link area, U∞ = 40 m/s).

(a) LG (b) Solid

Figure 9. Contours of velocity standard deviation in the xz-plane at y=0 (torque link area, U∞ = 40 m/s).

10 of 20

American Institute of Aeronautics and Astronautics

The wake structures of the solid and perforated fairings configurations are very similar. Toallow for a quantitative comparison between these configurations, the time averaged variablesof these configurations are subtracted in Figure 10. Due to the similarity in wake structure,

(a) Time averaged velocity (b) Standard deviation of velocity

Figure 10. Difference between solid and perforated fairings for contours in the xz-plane at y=0 (torque linkarea, U∞ = 40 m/s).

the direction of the subtracted vectors in Figure 10 aligns with the original flow direction inFigure 8(b). Downstream of the torque link area between z=100 and z=300 mm, wake velocitiesare up to 30 % lower for the perforated fairing. Plausibly, bleeding of air through the fairingsreduces the high speed deflection past the fairing sides and thereby reduces wakes velocities in thecenterplane. The earlier hypothesized large scale fluctuations associated with fairing size wouldthen also reduce in magnitude. This can explain the decrease in turbulence intensity reflected inthe standard deviation in Figure 10(b).

The velocity decrease in the torque link area is opposed to a velocity increase in the small jetbetween z=0 and z=50 mm. It is likely that the bled air through the undertray mixes with theflow that escapes from the undertray side edge and causes this velocity increase. Accompaniedwith the decrease of time averaged velocity comes a slight decrease in u′ as depicted in the lowerhalf of Figure 10(b).

Fluctuating surface pressures have been measured at several locations in the near wake (Figure2). The data for location 1 and 5 are shown in Figure 11. The spectra in the back of the mainleg are dominated by large scale turbulence and it’s breakdown. With both solid and perforatedfairings on, a hump is visible centered at f ≈ 104 Hz. The frequency scales with freestreamvelocity and using the main leg diameter, the non dimensional frequency combines to Str = 0.2.The difference between solid and perforated fairings is small, although against expectations theperforated fairing exhibits slightly higher amplitude fluctuations. Without the fairings on, thehump disappears and the magnitude of the pressure fluctuations is significantly lower. This iscombined with a 23% increase in base pressure to Cp = −0.79. Although the freestream flow candirectly impinge on the main leg at this z-coordinate, it seems that the vortex shedding is lessintense than with fairings on. The fact that the application of the fairings enhances the vortexshedding at this location probably lies in the overlay between the upper articulation link fairing

11 of 20

American Institute of Aeronautics and Astronautics

(a) Location 1, Main leg upper (b) Location 5, Rear bogie cap

Figure 11. PSD and mean surface pressure on the gear (U∞ = 40 m/s).

and the main leg, allowing the fairing boundary layer to flow past the sides of the main leg andseparate from it.

The sensor in the rear bogie cap shows the inverse image of the sensor in the upper main leg.The PIV-measurements revealed that shielding of the bogie beam by the undertray results in adead flow area at the rear of the bogie beam. Therefore fluctuations are of lower amplitude andbase pressure is higher compared to the configuration without fairings. The difference betweensolid and perforated fairings is small for both unsteady and steady pressures.

V.C. Wheel Flow

The flow around the landing gear wheels are investigated using (un)steady pressure measurements.A local polar coordinate system can be defined originating in the wheel axis to describe locationof the centerline tappings. The polar angle ϕ is defined in the inset in Figure 12 and starts at themost upstream wheel position.

V.C.1. Time averaged pressures

The centerline tappings show few difference between the configurations in Figure 12. The frontwheel exhibits separated flow roughly between ϕ = 90 ◦ and ϕ = 270 ◦, similar for all three configu-rations. The average level of the base pressure differs slightly, the LG configuration exhibiting thehighest value. The base pressure is not constant and shows a dip at ϕ = 190 ◦ with an amplitudeof about ∆Cp ≈ 0.3 for all configurations.

The rear wheel is showing a similar distribution to the front wheel in Figure 12(b). The 33 ◦

inclination of the bogie beam with respect to the freestream flow direction displaces the wheelsand prevents the wake of the front wheel from impinging directly on the rear wheel. There are fewdifferences between the configurations, apart from the LG configuration exhibiting a 35% higherbase pressure level. Base pressures are more constant in comparison to the front wheel. Also,suction is less on the lower wheel side (ϕ > 270 ◦) illustrated by a minimum value of Cp = −0.8opposed to Cp = −1.1 for the front wheel. Additionally, the point of minimum pressure on theupper wheel side shifts from ϕ = 90 ◦ to ϕ = 70 ◦, advancing the separation point.

The centerline mean pressure signatures of both front and rear wheels indicate that perforatingthe fairings is not significantly altering the time averaged flow around the wheels. The results fromthe shoulder taps confirm this image and are omitted here.

12 of 20

American Institute of Aeronautics and Astronautics

(a) Front wheel (b) Rear wheel

Figure 12. Centerline surface pressures on landing gear wheels (U∞ = 40 m/s).

V.C.2. Unsteady pressures

The PSD of the fluctuating pressures are shown in Figure 13 for several angular stations. Thelanding gear configurations without fairings, with perforated fairings and solid fairings are indi-cated using black, red and blue color respectively. The front and rear wheels can be distinguishedby a dashed and solid line respectively. For the attached flow regions at ϕ = 348 ◦ and ϕ = 303 ◦,the PSD exhibits a low value since large amplitude fluctuations are not present. There is very fewdifference between the various configurations and the front and rear wheel. A low frequency toneis present at f = 48 Hz. The spectral content of the tone is found to scale linear with velocity,implying the physical background is related to a shedding phenomenon. For the other angularstations, this low frequency tone is swamped by separation induced pressure fluctuations. Thehigh frequency content for the attached flow regions is dominated by peaks in the PSD due toelectronic noise.

For the front wheel, the unsteady pressures for the solid and perforated fairing at ϕ = 33 ◦

and ϕ = 78 ◦ show a similar signature to the PSD’s at ϕ = 348 ◦ and ϕ = 303 ◦. However,the rear wheel pressures show a low frequency hump centered roughly around f = 125 Hz, mostpronounced for the solid fairing configuration. Application of perforations reduces the peak levelof the fluctuations and shifts the center slightly upward to f = 150 Hz. The absence of thisphenomenon for the front wheel and the proximity of the angular stations to the front wheelpoints towards the unsteady front wheel wake responsible for the pressure fluctuations. However,the fact that the low frequency hump is hardly there for the plain landing gear configurationcould also point in the direction of the undertray side edge vortex (visualized as nr.3 in Figure 7)impinging on the rear wheel tyre surface. The linear scaling of the humps with freestream velocityconfirms the physical background of the phenomenon.

The high level of the PSD’s for the rear wheel at ϕ = 78 ◦ agrees with the forward movement ofthe separation point between the front and rear wheel as measured in the centerline mean pressuredistribution in Figure 12. The stations at ϕ = 123 ◦, ϕ = 168 ◦ and ϕ = 213 ◦ are located in theseparated flow region, reflected by the level increase of the PSD. The fairing configurations appearto exhibit more unsteadiness than the landing gear configuration. The difference between porousor solid fairings is small. At ϕ = 258 ◦, the flow is on the verge of separation at the undersideof both rear and front wheel, explaining the level decrease of the PSD. As for the attached flowregions at ϕ = 348 ◦ and ϕ = 303 ◦, the differences between configurations are small here.

Averaging the unsteady pressures yields the mean pressures around the wheel circumference.

13 of 20

American Institute of Aeronautics and Astronautics

(a) ϕ = 33 ◦ (b) ϕ = 78 ◦

(c) ϕ = 123 ◦ (d) ϕ = 168 ◦

(e) ϕ = 213 ◦ (f) ϕ = 258 ◦

(g) ϕ = 303 ◦ (h) ϕ = 348 ◦

Figure 13. PSD of surface pressure on landing gear wheels (U∞ = 40 m/s).

14 of 20

American Institute of Aeronautics and Astronautics

The obtained centerline pressure distributions in steps of ∆ϕ = 45 ◦ largely agree with the cen-terline pressure distributions as measured by the tappings discussed in section V.C.1.

VI. Acoustics

The noise of the landing gear model has been investigated using microphone arrays and on-surface microphones.

VI.A. Microphone array

Beamforming plots centered at f = 2 kHz for the different configurations are shown in Figure 14.The results reveal a decrease in noise source strength for this particular frequency by application

(a) (b) (c)

(d) LG (e) Solid (f) Perf

Figure 14. Averaged beamforming plots centred at 2 kHz from ceiling array (above) and side wall array(below) (U∞ = 40 m/s).

of perforated fairings. From the ceiling view, the main source is located between the wheels andremains there for the different configurations. The ground view from the side wall array shows thenoise source reduce in size and move downstream with solid fairings applied. The source furtherdiminishes in strength by application of the perforations.

At higher frequencies the resolution becomes better. Figure 15 shows clearly discernible sourcescentered at 10 kHz. From the ceiling view perspective, the solid fairing configuration is themost quiet for this frequency, showing an almost uniform background noise level. Application ofperforations gives perforate noise on the upper articulation link fairing, elevating the backgroundnoise level considerably. Without fairings a noise source emerges between upper articulationlink and the main leg. Additionally, a noise source appears at the brake rod junction with theyoke, plausibly because this item is not shielded anymore by the undertray. The location isconfirmed by the ground view localization plot for this configuration in Figure 15(d). Shieldedby the wheels for the ceiling view, noise sources emerge at the brake discs for the ground view.Application of fairings reduces the impingement velocity on the brake discs and hence the source

15 of 20

American Institute of Aeronautics and Astronautics

(a) (b) (c)

(d) LG (e) Solid (f) Perf

Figure 15. Averaged beamforming plots centred at 10 kHz from ceiling array (above) and side wall array(below) (U∞ = 40 m/s).

strength decreases, especially for the front breaks. The perforated fairings in Figure 15(f) showthe perforate noise again, together with the elevated background noise level. These results clearlyshow that especially for high frequencies, shielding sharp objects from high speed impingementyields considerable noise reduction.

In order to compare the results over the frequency domain, an area integration is performedover the different beamforming plots. For the ceiling array, values are added in the rectanglebetween x = ±0.2 m and z = −0.1 m to z = 0.3 m. For the side wall array, the rectanglecomprises of x = −0.4 m to x = 0.3 m and y = ±0.15 m. By covering various parts of theperforated fairings with tape, a study is performed on the effectiveness of perforation location.The undertray and articulation link stagnation area are deduced from the oil flow visualization inFigure 4.

A comparison showing the effect of the main configurations referenced to the solid fairingconfiguration is illustrated in Figures 16(a) and 16(b) for ceiling and side wall array respectively.The ceiling array shows a noise decrease by application of solid fairings in the mid frequencyrange and above f = 10 kHz. Lower frequencies (below f = 600 Hz) are noisier, but applicationof perforations makes this low frequency increase disappear. However, exposing all perforationsshows perforate noise emerging above f = 8 kHz. By exposing only the stagnation area of theperforations, the low frequency decrease persists while the perforate noise disappears. The side wallarray corresponding to the ground view shows a similar trend. The plain landing gear configurationis noisier above f = 1 kHz instead of the smaller frequency range for the ceiling array. The dip forall presented configurations at the f = 500 Hz band indicates a sharp noise increase in this bandfor the referenced solid fairing configuration. Although the limited resolution of the beamformingplots at this frequency hamper source localization, inspection of the plots reveals a large noisesource at the gear for the solid configuration opposed to wake noise sources for the plain landinggear and perforated fairing configuration.

16 of 20

American Institute of Aeronautics and Astronautics

(a) Effect of solid and perforated fairings (b) Effect of solid and perforated fairings

Ceiling array Wall array

Averaged between x = ±0.2 m and z = −0.1 m toz = 0.3 m

Averaged between x = −0.4 m to x = 0.3 m andy = ±0.15 m

Figure 16. Comparison of levels from integrated beamforming plots (referenced to Solid configuration,U∞ = 40 m/s).

VI.B. On-surface microphones

The narrowband spectra resulting from the on-surface microphone measurements for various lo-cations (Figure 2) are shown in Figure 17. The microphones in location 1 and 2 in the back ofthe strut are swamped by the large scale velocity fluctuations impinging on the back of the strut.Therefore acoustic phenomena are difficult to observe in these spectra and difference between con-figurations is hardly discernible. Unsteady pressure sensors are less sensitive than microphones,hence the PSD in Figure 11(a) for location 1 shows a different image to the narrowband spectrumin Figure 17(a).

At the lower main leg location, the LG configuration shows a hump at f ≈ 104 Hz or Str ≈ 0.2based on freestream velocity and leg diameter. It is believed to correspond to vortex sheddingnoise from the main leg. Since the fairings upstream of this part of the strut prevent high speedimpingement on the main leg, the hump is not visible with the fairings applied. Also, a peakcorresponding to perforate noise is visible just above f = 10 kHz with all perforations exposed.The peak scales linearly with freestream velocity U∞. The perforate noise is also picked up bythe microphones in locations 3 and 5. A localization study utilizing tape is carried out for themicrophone measurements. It reveals the sides of the lower articulation link being responsiblefor the perforate noise peak just above f = 10 kHz. This is conform to results in previousliterature,6 which already showed the shearing flow past the perforate is causing most of the noise.Comparing the non-dimensional frequency (referenced to local velocity and orifice diameter dor)with these experiments is difficult since the local velocity in the present experiment is unknown.Estimating this variable at 1.5 times the freestream velocity U∞ yields agreement between theStrouhal numbers for both experiments at Strdor = 0.33.

The spectra of the landing gear configuration at location 3 and 5 exhibit a higher level comparedto when the fairings are applied. Flow noise at the microphones is responsible for this higher level,since application of fairings prevents high speed flow in these areas. A clear difference betweensolid and perforated fairings is noticeable at location 3 above f = 100 Hz and almost up tof = 10 kHz. This elevated level of the solid fairing configuration in the low to mid frequencyrange agrees with the integrated wall- and ceiling array beamforming plot levels in Figure 16.Subtracting the tertsband averaged levels from this microphone further clarifies the differencebetween the configurations in Figure 18. A comparison to different freestream velocity values showsthe frequency content to scale linearly with velocity. Additionally, the difference in magnitude

17 of 20

American Institute of Aeronautics and Astronautics

(a) Location 1, Main leg upper (b) Location 2, Main leg lower

(c) Location 3, Bogie-articulation link junction (d) Location 5, Rear bogie cap

Figure 17. Narrowbandspectra of on surface microphones on the gear (U∞ = 40 m/s).

18 of 20

American Institute of Aeronautics and Astronautics

Figure 18. Subtracted tertsband averaged levels from microphone at bogie-artlink junction (referenced toSolid configuration).

between both configuration increases with freestream velocity. Interaction of the separated flowfrom the fairings with the downstream gear components is believed to be responsible for theelevated noise level above f = 100 Hz. In this frequency range, the solid configuration displaysa peak at f = 560 Hz which is believed to correspond to the dip in the integrated side wallarray levels in Figure 16. The perforation location effectiveness study utilizing tape shows thearticulation link fairing responsible for the noise. However, the peak stands out better in thefarfield wall array measurements than the nearfield on-surface microphone results.

Above f = 1000 Hz, the spectra in the dead flow areas (location 3 and 5) are far from smoothand electronic noise is dominant. Hence the several peaks in this part of the narrowband spectrashow no scaling with freestream velocity and are not believed to represent physical phenomena.The difference between all perforations exposed or only the stagnation point area perforations ishardly noticeable, except from the occurrence of the perforate noise above f = 10 kHz.

VII. Conclusion

An aerodynamic and acoustic survey has been performed on a simplified A340 look alike MainLanding Gear to explore the influence of (perforated) fairings on the lower part of the gear.

The oil flow visualization combined with the PIV have clarified flow direction and magnitudearound the fairings and in the wake centerplane. The undertray fairing exhibits largely attachedflow and divides the airstream in up- and downward direction. The downward pushed flow partlyescapes between front and aft wheels and produces a side edge vortex on the undertray side.The shielding of the bogie beam results in a dead flow area directly aft of the rear bogie end.The upward directed flow forms a horseshoe vortex around the lower articulation link fairingand pushes itself further upward on the upper articulation link fairing. The application of thefairings increases the frontal area and therefore bends the flow in transverse and vertical direction,preventing the airstream from penetrating in streamwise direction past the gear components. Thisinfluences the wake structure and hence the turbulence levels in the proximity of the downstreamgear components. The application of the perforations does not significantly alter the flow structurein the wake or in front of the fairings, indicated by equal flow directionality between solid andperforated configurations. However, flow velocity and turbulence levels in the near wake of thetorque link are lower than for impermeable fairings. It is hypothesized that by bleeding air, largescale unsteadiness associated with the fairings decreases resulting in this reduction.

The flow around the wheels is studied by means of centerline and shoulder pressure taps on thetyre surface. The mean surface pressures show few difference between the various configurations,

19 of 20

American Institute of Aeronautics and Astronautics

indicating that the time averaged flow around the wheels is largely unaltered by application of(perforated) fairings. The unsteady surface pressures show increased unsteadiness on the upperside of the rear wheels with fairings applied, pointing towards the undertray side edge vortexand/or the front wheel wake impinging on the rear tyre surface.

Upward from the torque link, unsteady pressure sensor measurements indicate that applicationof the fairings enhances the vortex shedding from the main leg at this position. By engineeringjudgement the overlay between upper articulation link fairing and the main leg is responsiblefor the enhanced shedding. Downward from this position the fairings prevent high velocitiesimpinging on the main leg. Therefore vortex shedding features at this position of the main legare only observed without fairings applied. The aerodynamic tests indicate that more porosity isneeded to induce more significant changes to the flow field by perforating the fairings.

The acoustics are studied employing several on-surface microphones and two microphone arraysto measure the ground- and side view noise signature. For both directions, application of solidfairings has been shown to reduce the noise in the mid and high frequency domain compared tothe plain landing gear configuration up to 4.5 dB. However a noise increase is measured in the lowfrequency domain. The application of perforations reduces the low frequency noise introduced bythe solid fairings to values below the plain landing gear configuration for both arrays. Additionally,reduced levels are measured in the mid frequency domain by application of the perforations. Thelow to mid frequency noise reduction is confirmed by measurements of an on-surface microphonelocated in the bogie beam - articulation link junction. Combined with the linear velocity scalingof the noise, this points in the direction of large scale separation of the fairings interacting withthe downstream gear components responsible for this phenomenon.

To investigate the effectiveness of perforation location, various parts of the fairing surfaceshave been covered with tape. It appears that the stagnation area perforations are responsible formost of the noise decrease, for both the articulation link and undertray fairing. Agreeing withbasic experiment results, most of the perforate noise is emitted from the lower articulation linksides at a Strouhal number of Str = 0.4 based on orifice diameter dor and local velocity .

VIII. Acknowledgements

Special thanks go to Matthew Spiteri and Mike Tudor-Pole for their assistance during the windtunnel experiments. Their support is greatly appreciated.

References

1Dobrzynski, W. and Buchholz, H., “Full-Scale Noise Testing on Airbus Landing Gears in the German DutchWind Tunnel,” AIAA Paper 97–1597, 1997.

2Dobrzynski, W., Chow, L., Guion, P., and Shiells, D., “A European Study on Landing Gear Airframe NoiseSources,” AIAA Paper 2000–1971, 2000.

3Piet, J.-F., Chow, L., Laporte, F., and Remy, H., “Flight Test Investigation of High-Lift Devices and LandingGear Modifications to Achieve Airframe Noise Reduction,” ECCOMAS 2004 conference, Jyvaeskylae, Finland, July2004.

4Smith, M., Fenech, B., Chow, L., Molin, N., Dobrzynski, W., and Seror, C., “Control of Noise Sources onAircraft Landing Gear Bogies,” AIAA Paper 2006–2626, 2006.

5Li, Y., Smith, M., Zhang, X., and Molin, N., “Noise Sources Control of an Aircraft Landing Gear,” AIAAPaper 2007–3465, 2007.

6Boorsma, K., Zhang, X., and Molin, N., “Perforated Farings for Bluff Body Noise Control,” AIAA Paper2007–3462, 2007.

7Smits, A. and Lim, T., Flow visualization: techniques and examples, Imperial College Press, 2000.8Raffel, M., Willert, C., and Kompenhans, J., Particle Image Velocimetry, A Practical Guide, Springer-Verlag,

corrected 3rd printing ed., 2000.9Moffat, R., “Describing the Uncertainties in Experimental Results,” Experimental Thermal and Fluid Sci-

ence, Vol. 1, 1988, pp. 3–17.10Fenech, B. and Takeda, K., “SotonArray: Southampton University Wind Tunnel Microphone Array System

Guide,” Tech. Rep. AFM 07/03, University of Southampton, School of Engineering Sciences, 2007.

20 of 20

American Institute of Aeronautics and Astronautics